ClawHub
Back to skill

Security audit

Knowledge Graph Buil

Security checks for vulnerabilities and agentic risk

Overview

This skill is not clearly malicious, but it asks for broad command and file access while describing callbacks, automatic cross-skill events, and external sync without enough scoping or enforcement.

Review this before installing if you use private project, team, Jira, CI/CD, or knowledge-base data. Keep callbacks and sync targets explicitly user-approved, prefer local-only graph operations unless you intentionally need integrations, and verify any kg commands before allowing snapshot restore, bidirectional sync, or automatic event subscriptions.

SkillSpector

Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Intent-Code Divergence

Medium
Confidence
89% confidence
Finding
The skill advertises network features such as callback URLs, cross-skill communication, and external integrations, while its security section claims HTTPS and certificate validation are enforced without showing any actual constraint or validation mechanism. This mismatch can cause downstream agents or operators to trust insecure callback or integration endpoints, enabling data exfiltration or man-in-the-middle exposure if plain HTTP or unvalidated TLS is used.

Description-Behavior Mismatch

Medium
Confidence
86% confidence
Finding
The metadata and description frame the skill as a knowledge-graph and local storage tool, but the document also introduces API calls, event bus communication, CI/CD hooks, Jira sync, and callback-based external communication. This scope expansion is dangerous because it can cause the skill to be granted broader trust and permissions than intended, increasing the chance that a caller unknowingly enables outbound communication or data transfer.

Vague Triggers

Medium
Confidence
83% confidence
Finding
The invocation description is broad and ambiguously scoped, covering database operations, SQL queries, storage management, visualization, versioning, cross-skill messaging, and multi-platform integration. Ambiguous scope is dangerous in an agent setting because it encourages overbroad invocation and tool use, which can lead to unnecessary file, shell, or network actions on data that the caller did not intend to expose or modify.

Static analysis

No suspicious patterns detected.